Formerly, most magnitude and phase vector measurements at microwave frequencies have been performed by network analyzers using techniques such as those described in "Automatic Network Analyzer 8542A, Section IV Network Analyzer Fundamentals", Hewlett-Packard Co. 1969 and in U.S. Pat. No. 4,244,024 issued Jan. 6, 1981 by Marzalek et al. Such vector network analyzers characterize networks, including devices, components, and systems by measuring the magnitude and phase of the network's transmission and reflection coefficients versus frequency. The capability to measure group delay, a special form of transmission and also usable in reflection, is also often incorporated in a vector network analyzer.
In general, a vector network analysis measurement system contains several separate modules. First is an RF source to provide the stimulus to the device under test (DUT). The stimulus normally covers a limited range of frequencies, either in a continuous analog sweep, referred to as the swept mode, in discrete steps, referred to as the step mode, or a single point mode. Second is a signal separation network to route the stimulus to the DUT and provide a means for sampling the energy that is reflected from, or transmitted through, the DUT. Also, energy is sampled from the signal that is incident upon the DUT in order to provide a reference for all relative measurements. Third is a tuned receiver to convert the resulting signals to an intermediate frequency (IF) for further processing. The magnitude and phase relationships of the original signals must be maintained through the frequency conversion to IF to provide usuable measurements. Fourth is a detector to detect the magnitude and phase characteristics of the IF signals, and fifth is a display on which to present the measurement results.
To improve measurement accuracy, a set of "standard" devices with known characteristics can be measured by a computer controlled system. From this data, a set of complex equations can be solved to determine a model representing many of the errors associated with the network analyzer process. This model is then stored in the computer and later when unknown devices are measured, the model can be used to separate the actual data from the "raw" measured data to provide enhanced accuracy in the microwave measurement by a process known as vector error correction.
Accuracy enhancement is very important in microwave measurements because even with the best signal generating and separating devices manufactured to state of the art tolerances, relatively large errors still occur as compared to low frequency measurements. For example, without vector error correction, a typical vector measuring system will yield errors of 30 percent. If one is willing to forego either the phase or impedance measurement of the unknown device, a modern scaler network analyzer is still only able to reduce the errors to 10 percent. On the other hand with prior implementations of vector error correction, errors can be reduced to about one percent.
Unfortunately, several significant problems remain with prior "automatic" network analyzers: they are very slow in the error correction mode; the systems are often quite awkward to use; they are unable to automatically perform a fully error-corrected measurement of forward and reverse reflection and transmission parameters (e.g., S.sub.11, S.sub.12, S.sub.21, and S.sub.22); and, broadband vector testing from RF to millimeter bands (e.g., 45 MHz to 26.5 GHz) cannot be performed with high accuracy and resolution without multiple manual reconnections.
Finally, in prior systems the data is usually displayed and analyzed only in the frequency domain, requiring either the use of a separate instrument such as a time domain reflectometer (TDR) in order to directly measure the response of the DUT as a function of time, or a powerful external computer coupled to the network analyzer to take data in the frequency domain and then perform an inverse Fourier conversion using either a truncated Fourier series or the faster Cooley-Tukey algorithm or others. Although the traditional TDR approach is fairly fast, the signal to noise ratio is low and the method is susceptible to both jitter and baseline drift. Conversely, although former computer coupled network analyzers exhibit significant improvements over the TDR method in signal to noise ratio, jitter, and drift, these systems are very slow, requiring several minutes to provide a time domain analysis and display of a DUT.